Presently, in Third Generation Partnership Project Radio Access Network Long Term Evolution (3GPP RAN LTE), an uplink sounding reference signal (SRS) is studied. Here, “sounding” refers to channel quality estimation and an SRS is mainly subject to time-multiplexing and transmitted in a specific time slot in order to estimate a CQI (Channel Quality Indicator) of an uplink data channel and estimate timing offset between a base station and a mobile station.
Further, possible methods of transmitting an SRS include the method of transmitting an SRS in a specific time slot in wideband and estimating a CQI over wideband at a time, and the method of transmitting a narrowband SRS in a plurality of time slots with shifting frequency bands (frequency hopping) and estimating a CQI over wideband in several times.
Generally, a UE (User Equipment) located near a cell boundary has significant path loss and a limitation of maximum transmission power. Accordingly, if an SRS is transmitted in a wideband, received power for a base station per unit frequency decreases and received SNR (Signal to Noise Ratio) decreases, and, as a result, the accuracy of CQI estimation deteriorates. Therefore, a UE near a cell boundary adopts a narrowband SRS transmission method of narrowing limited power to a predetermined frequency band and performing transmission. In contrast, a UE near the center of a cell has small path loss and received power for a base station per unit frequency can be kept enough, and therefore adopts a wideband SRS transmission method.
Meanwhile, another purpose of transmitting an SRS is to estimate timing offset between a base station and a mobile station. Accordingly, to secure the given accuracy of timing estimation Δt, the SRS bandwidth in one transmission unit (one frequency multiplexing unit) needs to be equal to or more than 1/Δt. That is, the bandwidth of an SRS in one transmission unit needs to fulfill both the accuracy of CQI estimation and the accuracy of timing estimation.
Further, in LTE, a PUCCH (Physical Uplink Control Channel), which is an uplink control channel, is frequency-multiplexed on both ends of the system band. Accordingly, an SRS is transmitted in the band subtracting the PUCCHs from the system bandwidth.
Further, the PUCCH transmission bandwidth (a multiple of the number of channels of one PUCCH bandwidth) varies according to the number of items of control data to be accommodated. That is, when the number of items of control data to be accommodated is small, the PUCCH transmission bandwidth becomes narrow (the number of channels becomes few) and, meanwhile, when the number of items of control data to be accommodated is great, the PUCCH transmission bandwidth becomes wide (the number of channels becomes large). Therefore, as shown in FIG. 1, when the PUCCH transmission bandwidth varies, the SRS transmission bandwidth also varies. In FIG. 1, the horizontal axis shows frequency domain, and the vertical axis shows time domain (same as below). In the following, the bandwidth of one channel of a PUCCH is simply referred to as the “PUCCH bandwidth” and the bandwidth by multiplying the PUCCH bandwidth by the number of channels is referred to as the “PUCCH transmission bandwidth.” Likewise, the bandwidth of an SRS in one transmission unit is simply referred to as the “SRS bandwidth” and the bandwidth of an SRS in plurality of transmission units is referred to as “SRS transmission bandwidth.”    Non-Patent Document 1: 3GPP R1-072229, Samsung, “Uplink channel sounding RS structure,” 7-11 May 2007